Add createLogger/trackDuration logging to mysql.ts, sql-server.ts, and
data-source.ts — the only database layer files without observability.
Connect/disconnect, query execution (with duration tracking), and
transaction lifecycle events are now logged. Passwords and parameter
values are excluded; SQL statements are capped at 100 chars.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Previously getLogDir() only checked app.isReady(), which caused
development builds to write logs to the user data directory instead
of the local project logs/ folder. Now uses app.isPackaged to
correctly distinguish production from development environments.
Also adds comprehensive logging system documentation and a debug
utility for verifying Electron environment detection.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add headless field to extraction config schema (default: true).
Extractor handler now reads globalConfig.extraction.headless instead
of hardcoding true. Users can set headless: false in config.yaml
to show the browser window during extraction for debugging.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove misleading await from audit-logger tests (functions are sync)
- Add cleanup() to LoggerAPI type definition in index.d.ts
- Fix circular reference fallback to preserve null values
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Make logAudit and closeAuditLogger synchronous (were async for no reason)
- Set audit-logger silent:true initially, enable on applyAuditConfig()
- Add try-catch for circular references in consoleFormat meta JSON
- Update all callers to remove unnecessary await/.catch() on sync functions
- Add comment to shared.ts explaining acceptable sync FS usage
- Fix audit-logger test for sync closeAuditLogger
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add LOGGER_LEVEL_CHANGED IPC channel for broadcasting level changes
- setLogLevel() now notifies all BrowserWindows when level changes
- Add verbose case in IPC forwardToWinston (was falling through to info)
- Renderer logger API listens for level changes and updates cached level
- Add cleanup() method to remove level change listener
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Cache isProduction() result at module load to avoid repeated property
lookups. Add isSerializedError() check in format functions to skip
re-serialization when error objects have already been processed by
logError/formatErrorForLogging.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
1. Extract shared module: consolidate getLogDir() and isProduction()
into shared.ts, eliminate duplication across logger modules
2. Make retention config effective: delay file transport creation
until config is loaded, apply appRetention/auditRetention from config.yaml
3. Add before-quit log flush: close logger and audit logger on
app exit to prevent log loss
4. Unify logError entry point: remove duplicate logError from index.ts,
re-export from error-utils.ts with richer error context
5. Renderer log level filtering: add client-side level check in preload
to skip IPC for filtered-out messages
6. Child logger cache + audit cleanup: cache child loggers in IPC
handler for performance, remove redundant timestamp format in audit logger
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add Admin-only delete button in operation history modal
- Replace dropdown with multi-select chip filters for Admin users
- Support filtering by multiple usernames using IN clause
- Fix user state propagation by passing currentUser via props
- Change GetBatchesOptions.username to usernames (array)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The database stores time in UTC format, and the UI should display UTC
time without timezone conversion. Use getUTCXxx() methods instead of
getHours() to avoid adding 8-hour timezone offset.
Also extract common datetime formatting logic to reduce code duplication.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Previously updateBatchStatus wrote the batch-level total recordCount to
every row, causing the detail view to show misleading identical counts.
Now mergeFiles collects per-order material counts, the handler writes
each order's count individually via updateRecordStatus, and batch
aggregation uses SUM instead of MAX for accurate totals.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
MySQL binary protocol prepared statements (connection.execute()) do not
support ? placeholders in LIMIT/OFFSET clauses, causing "Incorrect
arguments to mysqld_stmt_execute". Embed validated integer values directly
for MySQL while keeping parameterized queries for SQL Server.
Also apply React best practices to ExtractorOperationHistoryModal:
- Hoist formatDateTime to module level
- Wrap async handlers with useCallback for stable effect dependencies
- Import shared types instead of duplicating definitions
- Use ternary for conditional rendering
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add a new operation history feature for the extractor module that tracks
all extraction operations with persistent database storage.
Features:
- Records extraction operations with batch tracking (UUID-based)
- Preserves production ID to order number mapping
- Shows batch statistics (orders, records, success/failure counts)
- Expandable details for each batch showing individual order records
- User-based permission: Admin sees all records, User sees own records only
- Delete functionality with permission validation
Database:
- New ExtractorOperationHistory table schema
- Supports both SQL Server and MySQL
- Indexed on BatchId, UserId, and OperationTime
Files:
- Add DAO class for history operations
- Add IPC handler with permission checks
- Add preload API wrapper
- Add React modal component with expandable batch details
- Integrate history recording into extractor handler
- Add operation history button to ExtractorPage
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Remove all references to 'Guest' user type from the codebase, simplifying the role system to only support 'Admin' and 'User' roles.
Changes:
- Update type definitions to exclude 'Guest' from UserType
- Remove isGuest() method from SessionManager
- Remove Guest-specific logic from update services
- Update all type assertions from 'Admin | User | Guest' to 'Admin | User'
- Remove Guest UI styling from UserSelectionDialog
- Replace Guest fallback with ValidationError in settings handler
Error handling:
- Zod schema now rejects 'Guest' as invalid user type
- TypeScript will fail compilation if 'Guest' is referenced
- Runtime errors occur if database contains Guest users
No database migration needed (confirmed: no Guest users exist)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Change update check and download to async execution, allowing immediate app entry for User users
- Updates run in background, showing update prompt when download completes
- Optimize login flow experience, eliminating blocking time caused by update downloads
- Add release notes for version 1.5.1
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Change the protected row number range from 7000-7999 to 2000-7999 to prevent deletion of materials in this broader range.
- Updated isMaterialDeletable() method logic
- Updated getSkipReason() error messages
- Updated test cases to reflect new range boundaries
- Updated documentation templates and error collection guide
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add QueryResultRow interface to represent query results with order numbers
- Add collectQueryResultRows() method to extract order numbers upfront before processing
- Add extractOrderNumberFromQueryRow() helper to parse order numbers from query result cells
- Process rows with order number context instead of just row indexes
- Use actual order numbers in error details instead of BATCH_ROW_X placeholders
- Pass expected order number to detail processing for better validation
- Fix retry success handling to properly update statistics when retries succeed
This change provides better error context by associating each processed row with its actual order number from the query results, improving traceability and debugging.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add case-insensitive comparison for production ID database queries:
- SQL Server: use COLLATE SQL_Latin1_General_CP1_CI_AS
- MySQL: use UPPER() function for both field and input
- Map lookup: store keys in lowercase for consistent matching
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- CleanerPage now uses updateProcessConcurrency instead of setProcessConcurrency
- This ensures slider changes are persisted to config.yaml via IPC
- Remove unused queryBatchSize and setProcessConcurrency from destructuring
- Add *.yaml text eol=lf rule to .gitattributes for consistent line endings
- Format cleaner.ts with Prettier (parameter and chain formatting)
- Format CleanerPage.tsx (JSX formatting)
- Format cleaner.test.ts (array formatting)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add retry logic with max 2 attempts per failed order
- Track retry statistics (retriedOrders, successfulRetries)
- Generate detailed retry report section in execution reports
- Display retry metrics in ExecutionReportDialog UI
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Fix getTableName to properly split schema_tablename format
- Convert productionContractData_26年压力表合同数据 to [productionContractData].[26年压力表合同数据]
- Extend productionId pattern to support 1-6 digit serial numbers
Co-Authored-By: Claude (glm-5) <noreply@anthropic.com>
Core improvements:
- Fix focus restoration timing by using queueMicrotask only (removed double-layer async)
- Add comprehensive error handling with dev-mode logging for all failure scenarios
- Validate element visibility (display: none, visibility: hidden) before restoring focus
- Check disabled state and implement fallback to nearest focusable ancestor
- Add findNearestFocusableElement() helper for robust fallback strategy
- Add tabindex="-1" to focusable selectors for better focus management
- Use preventScroll option when calling focus() to prevent scroll jumps
Additional fixes:
- Remove unnecessary type conversion in Modal.tsx
- Fix TypeScript unused variable errors in main process
- Clean up unused imports in bip-users-dao.ts
- Add ARIA attributes and focus management to LoginDialog
- Add triggerRef support to UserSelectionDialog and ExecutionReportDialog
- Refactor ExecutionReportDialog to use Modal component
- Improve MaterialTypeManagementDialog with focus management
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Add error-utils module with serializeError and sanitizeError utilities
- Enhance IPC error handling to capture full error context including stack traces
- Add logError helper function for consistent error logging across the application
- Update console and file log formats to properly serialize error objects
- Replace all basic error logging in BIPUsersDAO with structured logError calls
- Add ErrorLike and SerializedError type interfaces for type safety
This improves debugging capability by preserving full error details in development
while sanitizing sensitive information in production logs.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Fix all TS6133 errors (unused variables) across service layer:
- Remove unused imports (path, ExtractionProgress type)
- Prefix unused parameters with underscore (_session, _totalBatches, etc.)
- Remove unused _verbose field and constructor from ExcelParser
- Remove unused _importToDatabase method from ExtractorService
- Remove unused _importProgress variable
This ensures clean type checking and eliminates dead code.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
This commit integrates the logging infrastructure across the entire application:
IPC Layer:
- Add logger-handler.ts with centralized IPC logging channels
- Integrate audit logging into auth, cleaner, extractor handlers
- Add structured logging for IPC operations and data flow
Service Layer:
- Add logger integration to ERP services (extractor, cleaner)
- Integrate logging into excel-parser and user DAO
- Add operation tracking and error logging
Renderer Layer:
- Add useLogger hook for component-level logging
- Update App.tsx with session and user activity logging
- Enable frontend audit trail for critical actions
Testing:
- Add comprehensive IPC logging integration tests
- Enhance unit test coverage for logger and audit-logger
- Add end-to-end logging flow validation
Types:
- Update preload type definitions for logging APIs
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
BREAKING CHANGE: Application now uses config.yaml instead of .env files
## Changes:
- Remove dotenv dependency from package.json
- Update all services to use ConfigManager for configuration
- Update tests to use fixed credentials instead of env vars
- Delete obsolete config-manager.test.ts (used old .env API)
- Update documentation (README.md, CLAUDE.md) to reflect new config system
## Configuration Architecture:
- ConfigManager: Centralized YAML configuration with Zod validation
- config.yaml location:
- Development: Project root (easy to edit and version control)
- Production: User AppData (persists across updates)
- ERP credentials: Stored in database (dbo_BIPUsers) per user
- Other settings: Stored in config.yaml (database, paths, extraction, etc.)
## Files Modified:
- package.json: Removed dotenv dependency
- cleaner-handler.ts: Use ConfigManager.getDatabaseType()
- run-migration.ts: Read from config.yaml instead of .env
- All integration tests: Use fixed test credentials
- tests/setup.ts: Removed dotenv loading
- README.md, CLAUDE.md: Updated documentation
Migration is complete. Application no longer depends on .env files.
- Add waitForLoginResult() method using Promise.race to detect:
- Success: .nc-workbench-icon element visible
- Failure: '名称或密码错误' error text visible
- Force login: click confirm button and re-detect
- Extract timeout constants (PAGE_LOAD_TIMEOUT, LOGIN_RESULT_TIMEOUT, FORCE_LOGIN_TIMEOUT)
- Improve error handling with clear error messages
- Add unit tests for class structure verification
- Fix test setup for Electron app mock
Fixes: ERP login success/failure detection was ambiguous
Add waitForLoginResult() method with Promise.race to detect three login outcomes:
- Success: detects .nc-workbench-icon element
- Failure: detects '名称或密码错误' error text
- Force login: clicks confirm button and re-detects
Improves login reliability by properly handling all authentication scenarios.
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-opencode)
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
